With the developments of economy and society, more and more banknotes have been issued and the currency of the banknotes is more frequent. Like those products that are recycled and destroyed after a period of use due to quality degrading, banknotes are also recycled and destroyed if they are degraded in quality after a period of use. Especially in some center branches of banks where a huge amount of banknotes are dealt, in the case that the banknote should be recycled and destroyed due to the stain, it is a tremendous amount of work to select the stained banknotes to be recycled and destroyed from all the banknotes in accordance with their categories. If the task is done manually, it is a laborious and time-consuming task which has a high cost. However, if the task is executed by a machine, the amount of manual labor will be saved, thus greatly reducing the cost of classifying stained banknotes.
In the prior art, a region of a paper on which a stain detection object is located is stored in advance as a detection object region, and the number of pixels of an image printed in the detection object region is stored in advance as a reference pixel number which is used for the comparison.
In practical applications, an aging degree of a paper-like medium to be detected may affect the determination of its stain depth, therefore if the stain determination is performed according to the prior art by using image pixels of a new paper-like medium as reference pixels, errors in the stain determination may be caused in the case where the paper-like medium to be detected is relatively old.